[PATCH] audit: we don't need to __set_current_state(TASK_RUNNING)

[PATCH] audit: we don't need to __set_current_state(TASK_RUNNING)

[Paul Moore]

From: Paul Moore <paul@paul-moore.com>

Remove the calls to __set_current_state() to mark the task as running
and do some related cleanup in wait_for_auditd() to limit the amount
of work we do when we aren't going to reschedule the current task.

Signed-off-by: Paul Moore <paul@paul-moore.com>
---
 kernel/audit.c |   12 +++++-------
 1 file changed, 5 insertions(+), 7 deletions(-)

diff --git a/kernel/audit.c b/kernel/audit.c
index 3a3e5de..f52fbef 100644
--- a/kernel/audit.c
+++ b/kernel/audit.c
@@ -430,7 +430,6 @@ restart:
 					attempts, audit_pid);
 				set_current_state(TASK_INTERRUPTIBLE);
 				schedule();
-				__set_current_state(TASK_RUNNING);
 				goto restart;
 			}
 		}
@@ -1324,15 +1323,14 @@ static inline void audit_get_stamp(struct audit_context *ctx,
 static long wait_for_auditd(long sleep_time)
 {
 	DECLARE_WAITQUEUE(wait, current);
-	set_current_state(TASK_UNINTERRUPTIBLE);
-	add_wait_queue_exclusive(&audit_backlog_wait, &wait);
 
 	if (audit_backlog_limit &&
-	    skb_queue_len(&audit_skb_queue) > audit_backlog_limit)
+	    skb_queue_len(&audit_skb_queue) > audit_backlog_limit) {
+		add_wait_queue_exclusive(&audit_backlog_wait, &wait);
+		set_current_state(TASK_UNINTERRUPTIBLE);
 		sleep_time = schedule_timeout(sleep_time);
-
-	__set_current_state(TASK_RUNNING);
-	remove_wait_queue(&audit_backlog_wait, &wait);
+		remove_wait_queue(&audit_backlog_wait, &wait);
+	}
 
 	return sleep_time;
 }

Re: [PATCH] audit: we don't need to __set_current_state(TASK_RUNNING)

[Richard Guy Briggs]

On 16/04/05, Paul Moore wrote:
> From: Paul Moore <paul@paul-moore.com>
> 
> Remove the calls to __set_current_state() to mark the task as running
> and do some related cleanup in wait_for_auditd() to limit the amount
> of work we do when we aren't going to reschedule the current task.

Wow, the logic of this goes back to 9ad9ad38 (2005-06-22, AUDIT: Wait
for backlog to clear when generating messages.)

> Signed-off-by: Paul Moore <paul@paul-moore.com>

These changes look good to me.

> ---
>  kernel/audit.c |   12 +++++-------
>  1 file changed, 5 insertions(+), 7 deletions(-)
> 
> diff --git a/kernel/audit.c b/kernel/audit.c
> index 3a3e5de..f52fbef 100644
> --- a/kernel/audit.c
> +++ b/kernel/audit.c
> @@ -430,7 +430,6 @@ restart:
>  					attempts, audit_pid);
>  				set_current_state(TASK_INTERRUPTIBLE);
>  				schedule();
> -				__set_current_state(TASK_RUNNING);
>  				goto restart;
>  			}
>  		}
> @@ -1324,15 +1323,14 @@ static inline void audit_get_stamp(struct audit_context *ctx,
>  static long wait_for_auditd(long sleep_time)
>  {
>  	DECLARE_WAITQUEUE(wait, current);
> -	set_current_state(TASK_UNINTERRUPTIBLE);
> -	add_wait_queue_exclusive(&audit_backlog_wait, &wait);
>  
>  	if (audit_backlog_limit &&
> -	    skb_queue_len(&audit_skb_queue) > audit_backlog_limit)
> +	    skb_queue_len(&audit_skb_queue) > audit_backlog_limit) {
> +		add_wait_queue_exclusive(&audit_backlog_wait, &wait);
> +		set_current_state(TASK_UNINTERRUPTIBLE);
>  		sleep_time = schedule_timeout(sleep_time);
> -
> -	__set_current_state(TASK_RUNNING);
> -	remove_wait_queue(&audit_backlog_wait, &wait);
> +		remove_wait_queue(&audit_backlog_wait, &wait);
> +	}
>  
>  	return sleep_time;
>  }

- RGB

--
Richard Guy Briggs <rbriggs@redhat.com>
Senior Software Engineer, Kernel Security, AMER ENG Base Operating Systems, Red Hat
Remote, Ottawa, Canada
Voice: +1.647.777.2635, Internal: (81) 32635, Alt: +1.613.693.0684x3545

Re: [PATCH] audit: we don't need to __set_current_state(TASK_RUNNING)

[Paul Moore]

On Wednesday, April 06, 2016 12:56:24 PM Richard Guy Briggs wrote:
> On 16/04/05, Paul Moore wrote:
> > From: Paul Moore <paul@paul-moore.com>
> > 
> > Remove the calls to __set_current_state() to mark the task as running
> > and do some related cleanup in wait_for_auditd() to limit the amount
> > of work we do when we aren't going to reschedule the current task.
> 
> Wow, the logic of this goes back to 9ad9ad38 (2005-06-22, AUDIT: Wait
> for backlog to clear when generating messages.)

Yeah, I didn't think much of it until Jiri submitted that last patch and 
figured I should check how we were doing handling the rescheduling in the rest 
of the code.  I think it stayed the way it was for so long because technically 
it wasn't hurting anything.

> > Signed-off-by: Paul Moore <paul@paul-moore.com>
> 
> These changes look good to me.
> 
> > ---
> > 
> >  kernel/audit.c |   12 +++++-------
> >  1 file changed, 5 insertions(+), 7 deletions(-)
> > 
> > diff --git a/kernel/audit.c b/kernel/audit.c
> > index 3a3e5de..f52fbef 100644
> > --- a/kernel/audit.c
> > +++ b/kernel/audit.c
> > 
> > @@ -430,7 +430,6 @@ restart:
> >  					attempts, audit_pid);
> >  				
> >  				set_current_state(TASK_INTERRUPTIBLE);
> >  				schedule();
> > 
> > -				__set_current_state(TASK_RUNNING);
> > 
> >  				goto restart;
> >  			
> >  			}
> >  		
> >  		}
> > 
> > @@ -1324,15 +1323,14 @@ static inline void audit_get_stamp(struct
> > audit_context *ctx,> 
> >  static long wait_for_auditd(long sleep_time)
> >  {
> >  
> >  	DECLARE_WAITQUEUE(wait, current);
> > 
> > -	set_current_state(TASK_UNINTERRUPTIBLE);
> > -	add_wait_queue_exclusive(&audit_backlog_wait, &wait);
> > 
> >  	if (audit_backlog_limit &&
> > 
> > -	    skb_queue_len(&audit_skb_queue) > audit_backlog_limit)
> > +	    skb_queue_len(&audit_skb_queue) > audit_backlog_limit) {
> > +		add_wait_queue_exclusive(&audit_backlog_wait, &wait);
> > +		set_current_state(TASK_UNINTERRUPTIBLE);
> > 
> >  		sleep_time = schedule_timeout(sleep_time);
> > 
> > -
> > -	__set_current_state(TASK_RUNNING);
> > -	remove_wait_queue(&audit_backlog_wait, &wait);
> > +		remove_wait_queue(&audit_backlog_wait, &wait);
> > +	}
> > 
> >  	return sleep_time;
> >  
> >  }
> 
> - RGB
> 
> --
> Richard Guy Briggs <rbriggs@redhat.com>
> Senior Software Engineer, Kernel Security, AMER ENG Base Operating Systems,
> Red Hat Remote, Ottawa, Canada
> Voice: +1.647.777.2635, Internal: (81) 32635, Alt: +1.613.693.0684x3545

-- 
paul moore
security @ redhat